Aims: To support a national approach to oxygen systems in Papua New Guinea, we conducted a study to document the incidence of hypoxaemia, its geographical distribution, epidemiological determinants and resource availability in several regions of the country. We also established baseline mortality rate data for all children admitted to five hospitals, for children with a diagnosis of pneumonia and for neonates to evaluate a future intervention.

Methods: Data were collected prospectively from over 1300 hospital admissions in five hospitals in 2004. To establish the baseline case fatality rates, data on outcome were collected retrospectively over 3 years (2001–2003) for over 20,000 children admitted to five hospitals.

Results: A total of 1313 admissions were studied prospectively in the five hospitals. Altogether, 384 (29.25%, 95% CI 26.8–31.8) had hypoxaemia, defined as SpO₂ <90%. The incidence of hypoxaemia was much greater in highland hospitals (40% of all admissions) than on the coast (10% of all admissions). This large difference in incidence persisted when the uniform definition of hypoxaemia was adjusted for altitude, and was largely because of differences in the incidence of acute respiratory tract infection. Oxygen was not available on the day of admission for 22% of children (range between hospitals, 3–38), including 13% of all children with hypoxaemia. Oxygen was less available in remote rural district hospitals than in provincial hospitals in regional towns. Clinical signs proposed by WHO as indicators for oxygen would have missed 29% of children with hypoxaemia and, if these clinical signs were used, 30% of children without hypoxaemia would have been considered in need of supplemental oxygen.

Conclusions: Based on this study, an approach to improving the detection of hypoxaemia and the availability of oxygen has been trialled in these five hospitals where a programme of clinical and technical training in the use and maintenance of pulse oximetry and oxygen concentrators has been introduced.